Ah -- I think the line from MIT Admissions at the time was that everyone was equally qualified, but they recruited more heavily in underrepresented groups. I didn't put much thought into exactly what measures they would take.
If you did take Bell Curve as completely true (which is a very lively debate), a shifted normal distribution would substantially change the makeup of a career field (otherwise equally distributed) picked from those with IQ>130 or something. It probably is fair to say Cisco engineers are smarter than the US average, although probably less so than top startup founders. In real life hiring is not on a single metric, of course, particularly later in one's career. But, a one+ SD shift would lead to really different populations at 115, 130, 160 IQ, in addition to absurd outliers like Einstein. (Plus, there's plenty to call into question IQ and specific measures of IQ, like Feynmann's relatively average score. I personally think it's far more predictive in the ~50-115 range than anywhere else.)
(The reproductive-years issue does seem like a fundamental one, especially in a career where your first 10 years are just the start. Are there any good URLs or books on how accounting and law handle this?)
First, note I'm not endorsing the Bell Curve, but exploring the implications.
There isn't a consensus on sex-linked differences in IQ, but none of the realistic studies show anything close to a 1+ SD shift. You see results like a 0.33 SD shift or a 1 point narrower standard deviation. What you see on the SAT Math is a 0.3 shift in mean and a slightly narrower SD.
Also, as a practical matter, engineering is not a profession of people 2SD+ above the mean. It ranges from Texas Tech to MIT. Typical IQ's for engineers are estimated around 110-120. If we look at SAT Math scores, 600+ would be a reasonable estimate. At the ~600 level, there are about the same number of males and females, with the gap growing to about 30% females in the perfect-800 pool (3SD above the mean). In other words the observed disparities far outstrip what would be expected from standardized test scores. Even taking the studies more favorable to the aptitude argument, you'd have to have engineers at 150+ before the observed male-female ratio was consistent with what would be expected based on IQ scores alone (5:1).
And of course, this assumes the only relevant measure of engineering aptitude is SAT Math performance or IQ. In that sense it's probably an upper bound for the measure of engineering aptitude.
Yes, the whole discussion is predicated on if Bell Curve were accurate.
I'd gone back to race, not gender. I think I've seen studies which say African-Americans are as much as a sigma below general population of the US, and certain Jewish or Asian populations are a sigma above, which is 2 sigma net, which is HUGE. I don't know if I buy these studies, but to the extent that IQ measures scholastic aptitude and culture vs. innate genetic intelligence, it's possible.
Engineers at large companies are maybe 1SD above the mean; founders or "10x engineers" at startups are 2SD+.
So, a 2 sigma difference at the 1 and 2 sigma above mean levels would be huge, which is observed in the population of startup founders and famous startup engineers. But there are plenty of other explanations which would account for exactly the same observation even if there were zero difference on population "aptitude" -- it's just one plausible explanation.
(There's also the argument that Asian immigrants to the US are potentially the top of a 3 billion person set, and the total number of African-Americans is something like 30mm. But the Ashkenazi Jewish population and African-American populations are on the same order of magnitude in the US.)
The race issue... well it's just not my little pet issue, LOL. My point is that I see these arguments being brought up in the context of women in engineering, not just minorities, and the underlying math doesn't support the conclusions even if we use the studies that are more favorable to the point. The numbers I've seen Richard Flynn throw around are a 1:5.5 ratio of women to men at 155+ (almost 4SD). While 1:5.5 is just a little under the representation of women in engineering, the 155+ figure is far beyond what you'd find for a practicing engineer. It might characterize the set of engineering professors at top schools.
If you did take Bell Curve as completely true (which is a very lively debate), a shifted normal distribution would substantially change the makeup of a career field (otherwise equally distributed) picked from those with IQ>130 or something. It probably is fair to say Cisco engineers are smarter than the US average, although probably less so than top startup founders. In real life hiring is not on a single metric, of course, particularly later in one's career. But, a one+ SD shift would lead to really different populations at 115, 130, 160 IQ, in addition to absurd outliers like Einstein. (Plus, there's plenty to call into question IQ and specific measures of IQ, like Feynmann's relatively average score. I personally think it's far more predictive in the ~50-115 range than anywhere else.)
(The reproductive-years issue does seem like a fundamental one, especially in a career where your first 10 years are just the start. Are there any good URLs or books on how accounting and law handle this?)